As artificial intelligence technology advances, so does AI-generated art. Services like Dall-E 3 and Midjourney aren't meant to replace human artists and creators, but these programs can help spark inspiration, aiding professional creators and giving amateur artists an easy way to create custom, personalized art. AI image generators are constantly evolving and offer endless creative opportunities.
The AI image generators on this list have undergone thorough testing. CNET reviewers have spent months with these programs, generating hundreds of images and creating everything from cartoon rainbow safaris to dramatic sci-fi scenes and photorealistic stock imagery. At some point, every service on this list spits out a wonky or unusable image. The test of a truly superior AI image generator is how well-equipped it is to handle those quirks and remediate flaws. Editing tools and customization options are a big part of that, which is why we test those extensively. Privacy policies, including how generators potentially train on your data, are also important considerations when selecting the best services. Here's what made the cut.
Dall-E 3 by OpenAI is our best AI image generator pick. It can handle long, complex queries, gives you many editing and customization options and its unique conversational flow makes it easy to bring your artistic visions to life. If you don't want to sign up for the $20 per month ChatGPT Plus subscription, Leonardo AI offers great AI imagery in its free plan. Beginners will find Canva's user-friendly design easy and accessible to use, and professional creatives looking to experiment with AI art ought to try Adobe Firefly and its family of AI tools.
CNET takes a practical approach to reviewing AI image generators. Our goal is to determine how good it is relative to the competition and which purposes it serves best. To do that, we give the AI prompts based on real-world use cases, such as rendering in a particular style, combining elements into a single image and handling lengthier descriptions. We score the image generators on a 10-point scale that considers factors such as how well images match prompts, creativity of results and response speed. See how we test AI for more.
Accuracy is an important factor to consider when evaluating AI image generators. Accuracy in this case refers to how well a service matches your prompt and how clearly elements and details are rendered. For example, a service that closely matches the sci-fi/fantasy spaceship scene described in a prompt would be considered accurate, as would a service that generates a picture of a human with a clear expression.
On the flip side, hallucinations, or the way AI makes stuff up, are also important to consider. Hallucinations for AI images tend to be seen in the weird quirks or flaws that you didn't request, like people with three arms or disappearing elements. Creativity is an inherent necessity of an AI image generator, but hallucinations are obvious flubs, not whimsical or non-realistic style elements. All AI image generators run into these problems, but great AI services have them less infrequently and offer editing tools ready to fix them.
The number of clarifying prompts required indicates how much work you'll have to put into getting the image you want. If you can't follow up with an edit or additional request, that can be a red flag or annoyance to look out for. Generators that adhere closely to prompts and offer editing tools make it easier to bring your vision to life.
Response speed is also important to consider. Most image generators are pretty quick, under two minutes or so. Services that can generate images quicker than that, between 10-30 seconds, bring an edge to their user experience.
Midjourney is a solid option for an AI image generator, but it didn't make our top picks because it's currently only available on Discord, is paid-only and inconsistently matches prompts. Midjourney does offer nice upscaling or editing tools for individual images, but you'll have to use them often. It's also noteworthy that all your images will be public and accessible in an online gallery unless you create in stealth mode, which is only available in the more expensive Pro and Mega plans.
Google's ImageFX also did not make our top picks due to persistent inaccuracies and its "overbearing nannying," which results in rejecting innocuous prompts. All AI generators struggle with photorealistic faces and hands, but ImageFX also struggles with inanimate objects and logos. This wouldn't be a deal-breaker, except its editing and fine-tuning tools aren't up to par to make images usable.
